Economic Impacts of Thailand's Biofuel Subsidy

energies

Article Economic Impacts of Thailand’s Biofuel Subsidy Reallocation Using a Dynamic Computable General Equilibrium (CGE) Model

Korrakot Phomsoda 1 , Nattapong Puttanapong 2 and Mongkut Piantanakulchai 1,*

1 School of Civil Engineering and Technology, Sirindhorn International Institute of Technology, Thammasat University, Pathum Thani 12120, Thailand; [email protected] 2 Faculty of Economics, Thammasat University, Bangkok 10200, Thailand; [email protected] * Correspondence: [email protected]; Tel.: +66-2986-9009

Abstract: For two decades, the Thai government has been promoting ethanol and biodiesel consumption through tax measures and price subsidies. Although this policy has substantially increased the consumption and production of biofuels, there is concern regarding its future ﬁscal burden. Due to ﬁscal constraints, the Thai government has planned to completely terminate the biofuel subsidy by 2022. This study aims at examining the economy-wide impacts of removing the biofuel subsidy and also conducting simulations of alternative scenarios, i.e., improving the yield of energy crops and reallocating the burden to expand capital investment in energy crop plantations. A recursive dynamic computable general equilibrium (CGE) model was used as the main quantitative method to conduct four simulation scenarios. This model was validated by comparing the simulation results

with the actual 2015–2019 data and showed low values of root mean square error (RMSE). The simulation results indicate that solely terminating the price subsidy would lead to economy-wide

Citation: Phomsoda, K.; contraction. Meanwhile, eliminating the price subsidy along with inﬂuencing crop yield improve- Puttanapong, N.; Piantanakulchai, M. ment and expanding capital investment in energy crop plantations would lead to the lowest negative Economic Impacts of Thailand’s impacts. Therefore, the termination of the price subsidy should be simultaneously implemented with Biofuel Subsidy Reallocation Using a supply-side expansions. Dynamic Computable General Equilibrium (CGE) Model. Energies Keywords: computable general equilibrium; biofuel; subsidy; energy crops; investment 2021, 14, 2272. https:// doi.org/10.3390/en14082272

Academic Editor: Robert Lundmark 1. Introduction Given the scarcity of fossil energy resources, Thailand must import an enormous Received: 14 March 2021 amount of energy, i.e., over 60% of its total domestic energy consumption annually [1], Accepted: 14 April 2021 Published: 18 April 2021 which results in a substantial economic loss every year. Consequently, the Thai government has placed importance on renewable energy development to replace fossil energy, especially

Publisher’s Note: MDPI stays neutral biofuels used in the transportation sector. For over two decades, the government has driven with regard to jurisdictional claims in biofuel production and consumption via tax measures and pricing policy, especially biofuel published maps and institutional affil- subsidies using an oil fund [2]. The production of purified biofuel (ethanol and biodiesel) iations. is costly [3,4]; therefore, a subsidy aims at maintaining biofuel prices lower than fossil fuel prices. The government uses the subsidy to offset the costs of biofuel feedstocks through petroleum refineries to farmers. As a result, energy crop farmers are indirect beneficiaries of the oil fund [5]; however, to compensate for biofuel prices, the government must provide a considerable budget, which, inevitably, affects the liquidity of the oil fund. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. Consequently, the current government decided to revise and issue the new State This article is an open access article Oil Fund Act B.E. 2562 (2019), which distinctly determines the oil fund allocation to distributed under the terms and occasionally stabilize domestic fuel prices during unusual situations (e.g., oil price shocks conditions of the Creative Commons in the global market). Additionally, the government has planned to completely eliminate Attribution (CC BY) license (https:// biofuel subsidies by 2022 [6]. Removing the biofuel subsidy triggers multidimensional creativecommons.org/licenses/by/ concerns involving formulating an appropriate energy policy that would be effective 4.0/). after 2022.

Energies 2021, 14, 2272. https://doi.org/10.3390/en14082272 https://www.mdpi.com/journal/energies Energies 2021, 14, x FOR PEER REVIEW 2 of 23

in the global market). Additionally, the government has planned to completely eliminate biofuel subsidies by 2022 [6]. Removing the biofuel subsidy triggers multidimensional Energies 2021, 14, 2272 concerns involving formulating an appropriate energy policy that would be effective2 of after 21 2022. Figure 1 shows an example of the biofuel price structure of Thailand. The graph . indicatesFigure that1 shows four anbiofuel example types of have the biofuelbeen moderately price structure subsidized of Thailand.The biofuels The graph sold in indicatesThailand that are fourcategorized biofuel typesas biodiesel have been and moderately gasohol. There subsidized. are three The types biofuels of biodiesel, sold in Thailandnamely H are-Diesel categorized (mandatory as biodiesel), H-Diesel and B7 gasohol. (option), There and H are-Diesel three B20 types (option of biodiesel, for heavy namelytrucks) H-Diesel. Gasohol (mandatory), is classified H-Dieselas Gasohol B7 (option),91, Gasohol and 95 H-Diesel (E10), B20Gasohol (option 95 for(E20), heavy and trucks).Gasohol Gasohol 95 (E85) is [7] classified. Retail prices as Gasohol comprise 91, Gasoholex-refinery 95 or (E10), producer Gasohol prices, 95 (E20), excise and tax, Gasoholmunicipal 95 tax, (E85) oil [ 7fund]. Retail levy, prices energy comprise conservation ex-refinery fund, value or producer-added tax prices, (VAT) excise on the tax, oil municipalproducer tax,(VAT1), oil fund marketing levy, energy margins, conservation and value fund,-added value-added tax on the oil tax trader (VAT) (VAT2) on the. oilThe producernegative (VAT1),value of marketing the oil fund margins, means the and fuels value-added are subsidized, tax on including the oil trader Gasohol (VAT2). 95 (E20), The negativeGasohol value 95 (E85), of the H- oilDiesel, fund and means H-Diesel the fuels B20 are. Ifsubsidized, the subsidy including is removed, Gasohol the retail 95 (E20), prices Gasoholwould increase 95 (E85), correspondingly, H-Diesel, and H-Diesel therefore B20. inducing If the subsidy higher isproduction removed, thecosts, retail household prices wouldexpenditures, increase and correspondingly, other costs of thereforeeconomic inducingactivities.higher production costs, household expenditures,In general, and removing other costs the of economicbiofuel subsidy activities. would have both negative and positive economyIn general,-wide implications, removing the i.e biofuel., on the subsidy one hand, would removal have of both the subsidy negative would and positive decrease economy-widethe oil fund budget implications, for offsetting i.e., on biofuel the one pric hand,es, while, removal on ofthe the other subsidy hand, would the government decrease thewould oil fund experience budget fora budget offsetting savings, biofuel thereby prices, lo while,wering on the the fiscal other burden hand, the and/or government allowing would experience a budget savings, thereby lowering the fiscal burden and/or allowing alternative options for fiscal expenditures. Studying the effects caused by such policy alternative options for fiscal expenditures. Studying the effects caused by such policy change requires a structural model that can capture the economy-wide propagations and change requires a structural model that can capture the economy-wide propagations and ultimate impacts. A computable general equilibrium (CGE) model is a tool widely used ultimate impacts. A computable general equilibrium (CGE) model is a tool widely used for this purpose. A CGE model comprises the interrelationships among economic agents. for this purpose. A CGE model comprises the interrelationships among economic agents. Changes in some variables in the model can concurrently affect other agents in the Changes in some variables in the model can concurrently affect other agents in the economy. economy. Therefore, a CGE model is an appropriate approach for policy study [8], Therefore, a CGE model is an appropriate approach for policy study [8], especially the nationwideespecially the impacts nationwide of fuel subsidies.impacts of fuel subsidies.

Figure 1. Biofuel price structure with the oil subsidy shown in purple. Figure 1. Biofuel price structure with the oil subsidy shown in purple. Conventionally, investment is an essential factor that drives economic growth because Conventionally, investment is an essential factor that drives economic growth it leads to an increase in employment in the short term and improves productivity by because it leads to an increase in employment in the short term and improves productivity strengthening production sectors in the long term. For Thailand, the average TFP growth by strengthening production sectors in the long term. For Thailand, the average TFP (for 2015–2018) was 1.6% [9]. In addition, considering the agriculture sector, its average – . . TFPgrowth growth (for was2015 only2018) 0.68%was per1 6% annum[9] In [10 addition,]. Therefore, considering improving the agriculture the TFP growth sector, isits necessaryaverage TFP for Thailandgrowth was because only it0.68% would per signiﬁcantly annum [10] lead. Therefore, to an improved improving GDP the andTFP highergrowth population income. On the basis of the description above, in this study, we focus on three policy options, namely (1) elimination of biofuel subsidy, (2) reallocation of the biofuel subsidy to investments in energy crop plantations, and (3) TFP enhancement of energy crops. Therefore, to achieve the results of the counterfactual experiments of three policy regimes, in this study, Energies 2021, 14, 2272 3 of 21

we apply the CGE model to address the structure of the Thai economy and examine the impacts on the economic system. The rest of this paper is structured as follows: In Section2, we introduce the methodology of the model; in Section3, we present the results and discussion; and in Section4, we summarize the results and brieﬂy suggest policy recommendations.

2. Literature Review Many previous studies have used a CGE model to analyze reforming energy subsidy, and have revealed that imposing a price distortion mechanism can cause inefficient resource allocations [11,12]. A study regarding Malaysia stated that, on the one hand, removing energy subsidies would increase the real GDP and real investment and would also decrease total energy demand, which would be beneficial to the economy. On the other hand, removal of energy subsidies would lead to a decrease in household consumption and welfare. As a result, their CGE-based simulation result recommended that the Malaysian government redistribute the budget to mitigate household effects [13]. Similarly, a study in Yemen revealed that although subsidy reduction would lead to economic growth, it would induce poverty expansion, and authors therefore suggested the government should transfer income to critical economic activities, such as the poorest households and public investments to enhance productivity and eliminate harmful effects [14]. In China, removal of the fossil fuel subsidies could result in negative consequences on output, potential competitiveness, energy demand, emission, and economic growth [15]. Given that subsidy removal could induce positive and negative effects, policies should focus on reallocating resources from the price subsidy to alternative measures [16]. In particular, using a structural model to simulate all possible scenarios generated by packages of regimes is necessary for formulating the optimal energy policy [17]. In Thailand, the oil fund mainly receives income from an oil levy, which is subse- quently reallocated to subsidize domestic prices of various fuels. Regarding the fund’s sustainability, several CGE-based studies related to energy taxes and subsidies have been conducted. The Asian Development Bank (ADB) has recommended that the Thai government remove the subsidy from fossil fuels and use it sporadically for mitigating world oil price fluctuations [1]. Another study has stated that, although eliminating the oil fund would lead to a lower GDP in the short term, the GDP growth would recover in the long term [18]. This study also indicated that, although the oil fund has been used for the biofuel subsidy to promote green energy, it has ignited energy market intervention. The results of terminating the subsidy would be similar to an increase in tax. If the government removes the biofuel subsidy, it would lower the GDP, social welfare, and energy consumption [5]. The above-mentioned studies have only analyzed the effects of tax and subsidy reforms. Another perspective involves investigating the benefits of a subsidy swap. Given that the removal of biofuel subsidies would mean that the government would retain more savings, some studies have suggested that some of the increased government savings should be reallocated [19]. Specifically, instead of freezing additional government savings earned from removing subsidies, the government should allocate the budget to investments in essential projects or compensation for people who face adverse effects [20]. According to these suggestions, the swap concept recommends that the government reallocate fossil energy subsidies to clean energy investments to expand economic growth; however, the results from reallocations vary by the area of study. An Indonesian study reported that eliminating energy subsidies and transferring money to invest in infrastructure and the renewable energy industry would positively impact the economy [21]. As shown by a study in South Africa, a decrease in the fuel subsidy and reallocating budgets to support public transportation would positively impact the economy [22]. In the case of Iran, eliminating energy subsidies and redistributing the budget to households and production sectors would result in a decrease in the GDP and higher inflation [23]. Given the diversity of national impacts, incorporating a combination of regimes in Thailand’s policy on subsidy elimination is crucial to the country’s development progress. Energies 2021, 14, 2272 4 of 21

3. Methodologies The model utilized in this study was adapted from a standard recursive dynamics CGE model published by the Partnership for Economic Policy (PEP) [24]. The procedures for constructing this model to simulate the economy-wide impacts in Thailand are explained in the following subsections.

3.1. Database The data showed that the interactions among economic agents in the CGE model were the social accounting matrix (SAM). The SAM was an economic accounting of income and expenditure developed from the Input-Output (I/O) table, provided by Thailand’s Office of the National Economic and Social Development Council [25]. This study incorporated new sectors and commodities for the energy policy analysis by using databases from various sources of Thailand, such as the Office of Energy and Policy Planning (EPPO), Department of Alternative Energy Development and Efficiency (DEDE), Department of Energy Business (DEB), Bank of Thailand (BOT), the Office of Agricultural Economics (OAE), and the Thailand Research Fund (TRF) [7,26–30]. The new SAM was comprised of 35 production sectors, 42 commodities (listed in Table A1 of AppendixA), and 3 institutes, i.e., the household, the government, and the rest of the world. There were two industries that produce multiple commodities: the petroleum refinery and sugar production industries. The petroleum sector was split into petroleum products, i.e., liquefied petroleum gas, jet oil and kerosene, gasohol, biodiesel, fuel oil, and other petroleum products. Given the deceleration of gasoline production and consumption, in this study, gasoline was combined with gasohol E10, whereas gasohol E20 and gasohol E85 were combined as gasohol E20/E85. Similarly, the type of biodiesel typically relies on government determination. This study separated biodiesel into biodiesel (mandatory) and biodiesel (option). The mandatory biodiesel was B5, whereas the optional biodiesel was combined with B7 and B10. Sugar production typically produces sugar products and molasses. Alternatively, ethanol can be produced by two industries, namely molasses-based ethanol production and cassava-based ethanol production. Although ethanol production in Thailand can be produced directly from sugar cane, this rarely occurs because sugar cane is much more suited for food manufacturing than energy.

3.2. Model Structure Figure2 shows the nested production structure in the model. At the top level, the production is a ﬁxed proportion between the primary and the intermediate factor following the Leontief production function (LEO). At the next level, labor and capital are merged with constant elasticity of substitution (CES) (The values of elasticity of substitution parameters are exhibited in AppendixA). At the same level, the total intermediate commodity is the integration between mixed gasohol, mixed biodiesel, and other commodities using the LEO. At the lowest level, there are two combinations of CES: mixed biodiesel comprises mandatory biodiesel and optional biodiesel, while gasohol is the combination of gasohol (E10) and gasohol (E20/E85). Regarding the other relationships in the model, household consumption aimed to maximize utility subjected to budget constraints following a Stone–Geary utility function. The investment demand varied inversely with commodity price. International trading was the substitution between commodity produced domestically and exports with constant elasticity of transformation (CET) (The values of elasticity of transformation parameters are shown in AppendixA). Similarly, the optimal selection between an import and the domestic commodity was determined by a function of constant elasticity of substitution. Energies 2021, 14, x FOR PEER REVIEW 5 of 23

Energies 2021, 14, 2272 other commodities using the LEO. At the lowest level, there are two combinations of 5CES: of 21 mixed biodiesel comprises mandatory biodiesel and optional biodiesel, while gasohol is the combination of gasohol (E10) and gasohol (E20/E85).

Figure 2. Nested production structure structure..

3.3. DynamicsRegarding Assumption the other relationships in the model, household consumption aimed to maximizeThis studyutility emphasizedsubjected to abudget short-term constraints period following of ten years a Stone (2022–2031).–Geary utility As introduced function. Theby the investment Partnership demand for Economic varied inversely Policy (PEP) with [commodity24], the recursive-dynamic price. International mechanism trading was theapplied substitution to the model between because commodity it allowed produced the intertemporal domestically relationship and exports between with constant invest- elasticityment and of capital transformation stock. (Alternatively, (CET) (The values the intertemporal of elasticity mechanismof transformation of the CGEparameters model arecan shown be fully in dynamic. Appendix For A). example, Similarly, the the mathematical optimal selection specifications between canan import be extended and the to domesticinclude the commodity fully endogenized was determined processes by ofa function intertemporally of constant optimizing elasticity behaviors of substitution [31,32]).. Also, this specification enabled the values of variables jointly involved in the dynamic 3.3.adjustment Dynamics to Assumption be exogenously defined based on the reports and official projections. Hence, these parameters were exogenously assigned to increase over time (The values of these This study emphasized a short-term period of ten years (2022–2031). As introduced exogeneous variables are shown in AppendixA), i.e., population, export, total investment, by the Partnership for Economic Policy (PEP) [24], the recursive-dynamic mechanism was public investment, government expenditures, productivity, depreciation rate, and the elas- applied to the model because it allowed the intertemporal relationship between ticity of investment demand. The population growth was adapted from the Organization investment and capital stock. (Alternatively, the intertemporal mechanism of the CGE for Economic Co-operation and Development (OECD) and the International Labor Or- model can be fully dynamic. For example, the mathematical specifications can be ganization (ILO) [33]. The export, total investment, public investment, and government extended to include the fully endogenized processes of intertemporally optimizing expenditures were based on the national income account reported by the Office of the behaviors [31,32]). Also, this specification enabled the values of variables jointly involved National Economic and Social Development Council [34]. The productivity was averaged in the dynamic adjustment to be exogenously defined based on the reports and official from the total factor of productivity reported by the OIE [35]. The depreciation rate and projections. Hence, these parameters were exogenously assigned to increase over time the elasticity of investment demand were applied from the previous studies [36,37]. (The values of these exogeneous variables are shown in Appendix A), i.e., population, export,3.4. Closure total and investment, Solution public investment, government expenditures, productivity, depreciationThe relationship rate, and among the elasticity variables of in investment the CGE model demand was. The comprised population of more growth equations was adaptedthan variables. from the Therefore, Organization some for variables Economic were Co fixed-operation and denoted and Development as exogenous (OECD) variables and theto solve International unique solutions Labor Organization that satisfy the(ILO) Walras [33]. lawThe [38export,]. In thistotal study, investment, the exogenous public investment,variables were and government government expenditures, expenditures public were sector based investment, on the national total investment, income account capital reportedstock, minimum by the Office consumption, of the National labor supply, Economic stock and change, Social and Development world prices Council of imports [34]. The and productivityexports. The numerairewas averaged was from an exchange the total rate. factor The of scale, productivity share, and reported exponential by the parameters OIE [35]. Thewere depreciation calibrated, following rate and thethe methodselasticity introducedof investment in previous demand studies were applied [37]. The from general the previousalgebraic studies modeling [36,37]. system (GAMS) was used to solve equations for equilibrium using a constrained nonlinear system (CNS).

Energies 2021, 14, 2272 6 of 21

3.5. Scenarios The Thai government aims to remove the biofuel subsidy, which is an enormous burden on the oil fund. In this study, we attempt to examine new measures for increasing benefits that could oppose the negative effects caused by higher biofuel prices. The new measures emphasize TFP and investments in energy crop plantations, which are upstream of the biofuel supply chain. Figure3 illustrates a flowchart for the reallocation of biofuel subsidy. The government levies all petroleum taxes, including excise tax, municipal tax, oil fund, energy conservation, and two value-added taxes (refinery and oil trading). The oil fund has been used to offset the purified biofuel prices. In this study, if the subsidy is removed, the government receives more income and savings. Then, it can reallocate some of its budget to invest in energy crop plantations, as shown in Table1. All scenarios involved complete removal of the biofuel subsidy. Scenario A (SIM A) had no reallocation of biofuel subsidy to investment and no change in TFP. SIM B had no reallocations but enhanced TFP by 1%. SIM C reallocated 10% of the biofuel subsidy to investment in energy crop plantations (e.g., sugar cane, cassava, and oil palm), but it had no change in TFP. Finally, SIM D reallocated 10% of the biofuel subsidy to investment in energy crop plantations and enhanced TFP by 1%.

Table 1. The four scenarios used for simulations.

% Reallocation for Capital Terminating Biofuel % Improvement on Scenarios Investment in Energy Crop Subsidy TFP of Energy Crops Plantations SIM A +100% - - SIM B +100% - +1% SIM C +100% +10% - SIM D +100% +10% +1%

The percentage changes in TFP are based on a previous study that reported Thailand had an increase in TFP of agriculture by less than 1% [39]. In addition, given that there have been no previous studies on energy crop investments, this study applied a policy scenario simulation ranging from no reallocation to reallocation when the subsidy elimination is effective. To explore long-term impacts, the model used in this study was based on the recursive dynamic approach. Therefore, some parameters were calibrated to ensure that a macroeconomic simulation resulted from the model that was in accordance with Thailand’s ofﬁcial data.

3.6. Sensitivity Analysis Although the CGE model has the key advantage of simulating the economy-wide price endogenous adjustments, the simulation result is considerably governed by the coefficients of elasticity of substitution. The Monte Carlo simulation can be applied to the standard CGE model to perform a sensitivity analysis [40,41]. Hence, the Monte Carlo was conducted by repeatedly simulating the model with the randomized sets of coefficients of elasticity of substitution. In particular, the coefficients determining the substitution of (1) gasohol (E10) and gasohol (E20/85), (2) biodiesel B5 and biodiesel B7/B10, and (3) labor and capital were jointly randomized. Energies 2021, 14, 2272 7 of 21 Energies 2021, 14, x FOR PEER REVIEW 7 of 23

– – Figure 3. Recycle flow ﬂow chart of biofuel subsidy into supply supply-sided-sided expansion for Scenarios A A–DD (SIMs A A–D).D).

3.6.4. Results Sensitivity and Analysis Discussion AlthoughThis section the is CGE divided model into has three the subsections. key advantage First, of wesimulating present thethe reliabilityeconomy-wide of the pricemodel, endogenous and then in adjustments, the second and the the simulation third subsections, result is weconsiderably describe the governed macroeconomic by the coefficientsimpacts, as of well elasticity as impacts of substitution. at the sector The level, Monte respectively. Carlo simulation can be applied to the standard CGE model to perform a sensitivity analysis [40,41]. Hence, the Monte Carlo was conducted4.1. Model Validationby repeatedly simulating the model with the randomized sets of coefficients of elasticityTo ensure of substitution. the precision In particular, of the model, the co theefficients dynamic determining parameters th weree substitution repeatedly of cali- (1) gasoholbrated until (E10) the and percentage gasohol (E20/85), of root mean (2) biodiesel square error B5 (RMSE)and biodiesel of the modelB7/B10, and and the (3) ofﬁcial labor anddata capital were almost were jointly the same. randomized. Figure4 shows the RMSE of macroeconomic indices lower than 10%. The maximum RMSE was government consumption expenditures of 6.86%, whereas 4.the Results minimum and RMSEDiscussion was private consumption (PCON) of 2.36%. The rest of the indices, includingThis section gross domestic is divided product into three (GDP), subsections gross ﬁxed. First, capital we formation present the (GFCF), reliability import, of andthe model,export, and displayed then in RMSEs the second of 3.49%, and the 2.62%, third 5.97%, subsections, and 3.69%, we describe respectively. the macroeconomic Consequently, impacts,the model as waswell reliable as impacts and at stable the sector for forecasting level, respectively economic. variations together. 4.2. Macroeconomic Impacts 4.1. Model Validation 4.2.1. Overview of Economic Adjustment To ensure the precision of the model, the dynamic parameters were repeatedly calibratedThe key until macroeconomic the percentage indices of root presented mean square in this error study (RMSE) were of GDP, the consumermodel and price the index (CPI), PCON, and GFCF. Table2 shows a summary of the 10-year average change for official data were almost the same. Figure 4 shows the RMSE of macroeconomic indices each scenario. lower than 10% The maximum RMSE was government consumption expenditures of In SIM A, completely. removing the biofuel subsidy leads to an economic recession; in 6this.86%, scenario, whereas GDP, the PCON,minimum and RMSE GFCF was decrease private by 0.013%,consumption 0.416%, (PCON) and 0.035%, of 2.36% respectively,. The rest ofwhereas the indices, CPI increases including by 0.171%.gross domestic product (GDP), gross fixed capital formation (GFCF),In SIMimport, B, the and complete export, removal displayed of the RMSEs biofuel subsidyof 3.49%, together 2.62%, with 5.97%, 1% TFPand enhance- 3.69%, respectivelyment of energy. Consequently, crops leads the to bettermodel outcomes was reliable than and SIM stable A. The for GDP forecasting is zero whicheconomic is a variationshigher value together than in. SIM A (negative), whereas PCON and GFCF are better than in SIM A, with a decrease of 0.403% and 0.029%, respectively. Contrarily, CPI increases by 0.158%,

Energies 2021, 14, 2272 8 of 21

Energies 2021, 14, x FOR PEER REVIEWa small degree as compared with SIM A. This implies that enhancing the TFP of energy8 of 23

crops can partially mitigate the negative impacts caused by removing the biofuel subsidy.

Figure 4. Discrepancies between model model-generated-generated results results and and official ofﬁcial data data of the macroeconomic indices for the years 2015–2019.2015–2019.

4.2. MacroeconomicIn SIM C, a combination Impacts of completely removing the subsidy and reallocating 10% 4.2.1.of the Overview biofuel subsidy of Economic to invest Adjustment in energy crop plantations is similar to SIM B. The GDP, PCON, and GFCF are reduced by 0.001%, 0.403%, and 0.024%, respectively, whereas CPI The key macroeconomic indices presented in this study were GDP, consumer price increases by 0.155%. This implies that sharing the biofuel subsidy to invest in energy crop index (CPI), PCON, and GFCF Table 2 shows a summary of the 10 year average change plantations can compensate for the. loss of economic beneﬁt caused by- removing the biofuel forsubsidy. each scenario. InFinally, SIM A, in SIMcompletely D, enhancing removing TFP the by 1%biofuel and reallocatingsubsidy leads 10% to ofan the economic biofuel subsidyrecession; to inenergy this cropscenario, plantations GDP, resultsPCON, in theand best GFCF economic decrease growth. by 0 The.013%, GDP 0 rebounds.416%, and to positive0.035%, respectively,as 0.012%, whereas whereas CPI CPI still increases increases by 0 by.171% 0.143%.. PCON and GFCF decrease by 0.390% and 0.018%,In SIM respectively.B, the complete Therefore, removal the outcomesof the biofuel fromthis subsidy simulation together indicate with that 1% using TFP enhancementTFP improvement of energy and reallocationcrops leads to of better the biofuel outcomes subsidy than forSIM investment A. The GDP in is energy zero which crop isplantations a higher value are optimal than in policies SIM Afor (negative), mitigating whereas adverse PCON effects. and GFCF are better than in SIM A, with a decrease of 0.403% and 0.029%, respectively. Contrarily, CPI increases by 0.158%, a small degree as compared with SIM A. This implies that enhancing the TFP of energy crops can partially mitigate the negative impacts caused by removing the biofuel subsidy. In SIM C, a combination of completely removing the subsidy and reallocating 10% of the biofuel subsidy to invest in energy crop plantations is similar to SIM B. The GDP, PCON, and GFCF are reduced by 0.001%, 0.403%, and 0.024%, respectively, whereas CPI

Energies 2021, 14, 2272 9 of 21

Table 2. Impacts of scenario simulations on macroeconomic indices (% change on average during 2022–2031).

Scenario Indices SIM A SIM B SIM C SIM D Gross domestic product (GDP) −0.013 0.000 −0.001 0.012 Consumer price index (CPI) 0.171 0.158 0.155 0.143 Private consumption (PCON) −0.416 −0.403 −0.403 −0.390 Gross ﬁxed capital formation (GFCF) −0.035 −0.029 −0.024 −0.018 Source, authors’ simulation. Note that instead of reporting with monetary value, this study displays the results as the percentage change for all scenarios (SIMs A, B, C, and D) as compared with the baseline scenario (BAU).

Figure5 shows the adjustment of economic agents and firms. In SIM A, the removal of the biofuel subsidy leads to a higher fuel tax on the first path that the government would gain more income. Then, the government has an income for consumption of goods and services. The rest of the government consumption would be transferred for saving, which could be combined with the savings of a household and a foreigner to become the total investment. On another path, the removal of biofuel subsidy leads to an increase in biofuel prices. The higher biofuel prices then lead to higher production costs, and commodity prices increase correspondingly, and consequently, the CPI also increases. Due to higher production costs, the aggregate outputs would decrease, decreasing the value added for all sectors. This situation leads to lower household income and saving. When the household income reduces, the consumption is also eventually decreased. Additionally, the lower household saving also affects government income. The offset between an increase in government saving and a decrease in household saving remains positive for the total investment. In this study, these three institutes (i.e., household, government, and the world) drive the total investment that is fixed by the closure rule. Given that the government savings increase, the foreign savings would be adjusted to decrease. This implies that removing the biofuel subsidy would indirectly affect a decrease in foreign direct investment. When foreign savings are decreased, it leads to increasing the current account balance and trade balance. Finally, the GDP change comprises private consumption, government consumption, GFCF, stock changes, and trade balance. In SIM A, the negative impacts of the subsidy removal were similar to findings of previous studies [13,42–47]. In SIM B, although the economy endures the negative effects of removing the biofuel subsidy (SIM A), the TFP improvement increases the outputs of energy crop plantations. Therefore, it leads to lower prices of biofuel products and productions; however, removing the biofuel subsidy has a stronger influence than the TFP increment. Therefore, enhancing the TFP of energy crops by 1% could elevate the GDP by 0.013% as compared with SIM A. Similarly, in SIM C, reallocating the biofuel subsidy to invest 10% in energy crop plantations enhances the GDP by approximately 0.012% as compared with in SIM A. The potential of SIM C is similar to that of SIM B. The investment is to increase capital supply for production sectors. Such an increase leads to the depression of capital costs. Unlike improvement productivity (SIM B), it would increase the quantity of outputs by a larger amount. By comparing SIMs B and C, we can show that a 10% increase in investment is close to a 1% increase in the TFP on energy crops. In addition, SIM D is a combination between SIMs B and C. This scenario increases the capital supply and efficiency of the energy crop sector. The results indicated that a 1% increase in the TFP and a 10% investment in energy crop plantations can enhance economic growth via a GDP of 0.025% (0.012−(−0.013)) as compared with SIM A. In summary, the effects of removing the biofuel subsidy could be avoided by enhancing the TFP and reallocating the biofuel subsidy to an investment in energy crop plantations, Energies 2021, 14, 2272 10 of 21

Energies 2021, 14, x FOR PEER REVIEW 10 of 23 which are appropriate policies for developing economic and energy production without energy price distortion.

Figure 5. A flowchart ﬂowchart of the economy-wideeconomy-wide impacts impacts.. 4.2.2. Trends of Macroeconomic Changes In SIM B, although the economy endures the negative effects of removing the biofuel •subsidyGross (SIM Domestic A), the ProductTFP improvement increases the outputs of energy crop plantations. Therefore,Figure 6it shows leads theto trendslower ofprices the impacts of biofuel caused products by removing and biofuelproductions; subsidies however, incor- poratedremoving with theTFP biofuel progress subsidy and has reallocation a stronger for influence investment than in the energy TFP cropincrement plantations.. Therefore, The Energies 2021, 14, x FOR PEER REVIEW 11 of 23 resultsenhancing show the that TFP removing of energy biofuel crops subsidiesby 1% could (SIM elevate A) could the GDP affect by an 0 economic.013% as compared recession becausewith SIM the A. average GDP decreased by 0.013%; however, the GDP adjustment slowly proceededSimilarly, until in 2031, SIM when C, reallocating the GDP would the biofuel rebound subsidy to be zero. to invest The reason 10% in for energy this is crop that reasoneliminating for this biofuel is that subsidieseliminating affects biofuel higher subsidies biofuel affects prices higher and leads biofuel to higher prices productionand leads plantations enhances the GDP by approximately 0.012% as compared with in SIM A. The tocosts. higher Then, production the demand costs and. Then, supply the ofdemand aggregate and outputs supply alsoof aggregate decrease. Therefore,outputs also the potential of SIM C is similar to that of SIM B. The investment is to increase capital supply decreaselower outputs. Therefore, result the in alower lower outputs GDP; however, result in the a lower trend ofGDP; the however, GDP would the be trend to gradually of the for production sectors. Such an increase leads to the depression of capital costs. Unlike GDPincrease. would This be to increase gradually is in increase agreement. This with increase a previous is in agreement study [17 with], which a previous indicated study that improvement productivity (SIM B), it would increase the quantity of outputs by a larger [17],removingwhich subsidiesindicated wouldthat removing enhance subsidies the GDP would in the longenhance term. the GDP in the long term amount . By comparing SIMs B and C, we can show that a 10% increase in investment. is close to a 1% increase in the TFP on energy crops. In addition, SIM D is a combination between SIMs B and C. This scenario increases the capital supply and efficiency of the energy crop sector. The results indicated that a 1% increase in the TFP and a 10% investment in energy crop plantations can enhance economic growth via a GDP of 0.025% (0.012−(−0.013)) as compared with SIM A. In summary, the effects of removing the biofuel subsidy could be avoided by enhancing the TFP and reallocating the biofuel subsidy to an investment in energy crop plantations, which are appropriate policies for developing economic and energy production without energy price distortion.

4.2.2. Trends of Macroeconomic Changes

• Gross Domestic Product FigureFigure 6. 6.ProjectedProjected impacts impacts of four of fourscenarios scenarios (SIMs (SIMs A–D) A–D)on gross on domestic gross domestic product product during 2022 during– Figure 6 shows the trends of the impacts caused by removing biofuel subsidies 20312022–2031.. incorporated with TFP progress and reallocation for investment in energy crop plantationsTo enhance. The the results TFP ofshow energy that crops removing (SIM B) biofuel, although subsidies the average (SIM economicA) could affectgrowth an wouldeconomic be zero, recession it remains because better the than average SIM A GDP. During decreased 2022–2025, by 0 the.013%; GDP however, would be the lower GDP thanadjustment zero, but slowly during proceeded 2026–2031, until the GDP2031, would when reboundthe GDP to would be positive rebound. The to two be periods,zero. The therefore, are offset by each other, because increasing the TFP induces higher total outputs of energy crops and lower production costs. The lower costs of cultivating and harvesting energy crops lead to lower biofuel prices. Given the inverse variation between commodity price and investment demand, lower biofuel prices would lead to higher investment demand for biofuel and continue to trigger capital accumulation in the next year. This process results in increasing the GDP gradually and implies that enhancing the TFP of energy crops could compensate for the negative effects of terminating biofuel subsidies. The results are in line with [37], which stated that improving bioethanol cultivation efficiency enhances economic growth because efficiency would reduce cost per unit and produce more products. Likewise, reallocating the biofuel subsidy to invest in energy crop plantations (SIM C) affects the average GDP by a decrease of 0.001%, similar to SIM B, because the investments increase the capital of energy crop cultivation. It leads to more production and less unit cost of production in energy crops. When production outputs are increased, the GDP loss is reduced, which indicates that reallocation could also compensate for adverse effects caused by cutting off biofuel subsidies. This simulation result is in line with the finding obtained from an econometrical approach [48]. Finally, if the TFP improvement and investment in energy crop plantations were implemented simultaneously (SIM D), the GDP would recover faster than in SIM B or SIM C. The average GDP growth was positive as 0.012%, because both implementations have double influences on the economy. The TFP would drive productivity, while reallocation would increase the supply of energy crops. Therefore, the GDP would resume to be positive quickly by 2025. Additionally, the average GDP in SIM D is nearly twice that in SIM A. In summary, all scenarios initially impact GDP loss, but the GDP would gently increase until it becomes positive in the long term. In this case, the best choice is the combination of enhancing TFP and reallocation of biofuel subsidy. The government can completely remove the biofuel subsidy without economic contraction. In the same line of the previous literature [49], these obtained simulation results indicate that the policy

Energies 2021, 14, 2272 11 of 21

To enhance the TFP of energy crops (SIM B), although the average economic growth would be zero, it remains better than SIM A. During 2022–2025, the GDP would be lower than zero, but during 2026–2031, the GDP would rebound to be positive. The two periods, therefore, are offset by each other, because increasing the TFP induces higher total outputs of energy crops and lower production costs. The lower costs of cultivating and harvesting energy crops lead to lower biofuel prices. Given the inverse variation between commodity price and investment demand, lower biofuel prices would lead to higher investment demand for biofuel and continue to trigger capital accumulation in the next year. This process results in increasing the GDP gradually and implies that enhancing the TFP of energy crops could compensate for the negative effects of terminating biofuel subsidies. The results are in line with [37], which stated that improving bioethanol cultivation efficiency enhances economic growth because efficiency would reduce cost per unit and produce more products. Likewise, reallocating the biofuel subsidy to invest in energy crop plantations (SIM C) affects the average GDP by a decrease of 0.001%, similar to SIM B, because the investments increase the capital of energy crop cultivation. It leads to more production and less unit cost of production in energy crops. When production outputs are increased, the GDP loss is reduced, which indicates that reallocation could also compensate for adverse effects caused by cutting off biofuel subsidies. This simulation result is in line with the finding obtained from an econometrical approach [48]. Finally, if the TFP improvement and investment in energy crop plantations were implemented simultaneously (SIM D), the GDP would recover faster than in SIM B or SIM C. The average GDP growth was positive as 0.012%, because both implementations have double influences on the economy. The TFP would drive productivity, while reallocation would increase the supply of energy crops. Therefore, the GDP would resume to be positive quickly by 2025. Additionally, the average GDP in SIM D is nearly twice that in SIM A. In summary, all scenarios initially impact GDP loss, but the GDP would gently increase until it becomes positive in the long term. In this case, the best choice is the combination of enhancing TFP and reallocation of biofuel subsidy. The government can completely remove the biofuel subsidy without economic contraction. In the same line of the previous literature [49], these obtained simulation results indicate that the policy reform can lead to the diverse outcomes. The incomplete policy set (SIM A) can incur the negative effect, while the appropriate policy package (SIM D) can influence the sustainable growth path. • Consumer Price Index Figure7 displays the effects of all scenarios on the CPI. In contrast to the GDP, removal of the biofuel subsidy (SIM A) would increase the CPI thoroughly during 2022–2031. Given the higher prices of biofuel, the production sectors would then suffer higher production costs. This result differs from a previous study [18] that reported removal of the oil fund would lead to a decline in the CPI. In this study, at the top level of the production function, as mentioned in Section 3.2, the substitution between primary input and intermediate commodities is assumed to be fixed, and there is no replacement by each other. In addition, mandate and optional biodiesel are subsidized. As a result, there is no choice for people or industry uses, i.e., They must assume there would be higher prices of biodiesel. Unlike gasohol, currently, only gasohol E20 and E85 are still being subsidized. If the subsidies on gasohol E20 and E85 are eliminated, both fuel prices would increase automatically. Instead of using gasohol E20 and E85, people could also switch to use gasohol E10. In reality, biodiesel is a crucial fuel for the transportation and industry sectors. Hence, increasing biodiesel prices caused by terminating subsidies significantly affects the economy. In contrast, in SIM B, the CPI initially would move up from 2022 to 2028. Afterwards, it would diminish slightly and relatively move as constant. Similarly, the CPIs of both SIMs C and B behave similarly. From 2022 to 2027, the CPI would increase, but from 2027 to 2031, it would be constant. SIM D shows an increase in CPI for only three years (2022–2025), and then it would be stable and decelerate until 2031, because the TFP improvement and investments in energy crop plantations would increase production output and reduce the Energies 2021, 14, x FOR PEER REVIEW 12 of 23

reform can lead to the diverse outcomes. The incomplete policy set (SIM A) can incur the negative effect, while the appropriate policy package (SIM D) can influence the sustainable growth path. • Consumer Price Index Figure 7 displays the effects of all scenarios on the CPI. In contrast to the GDP, removal of the biofuel subsidy (SIM A) would increase the CPI thoroughly during 2022– 2031. Given the higher prices of biofuel, the production sectors would then suffer higher production costs. This result differs from a previous study [18] that reported removal of the oil fund would lead to a decline in the CPI. In this study, at the top level of the production function, as mentioned in Section 3.2, the substitution between primary input and intermediate commodities is assumed to be fixed, and there is no replacement by each other. In addition, mandate and optional biodiesel are subsidized. As a result, there is no choice for people or industry uses, i.e., They must assume there would be higher prices of Energies 2021, 14, 2272 biodiesel. Unlike gasohol, currently, only gasohol E20 and E85 are still being subsidized12 of 21. If the subsidies on gasohol E20 and E85 are eliminated, both fuel prices would increase automatically. Instead of using gasohol E20 and E85, people could also switch to use gasoholproducer E10 prices. In reality, of commodities. biodieselTherefore, is a crucial the fuel consumer for the prices transportation of ﬁnished and goods industry would sectorsdecrease. Hence, accordingly increasing and promotingbiodiesel prices the TFP caused and investment by terminating in energy subsidies crop plantations significantly by affectsreallocating the economy the biofuel. subsidy are appropriate regimes to reduce inﬂation.

FigureFigure 7. 7. ProjectedProjected impacts impacts of four scenarios (SIMs A–D)A–D) onon consumerconsumer priceprice indexindex during during 2022–2031. 2022– 2031. In summary (see Table2), during such a period, the CPI in SIM A would increase by 0.171%In contrast, on average, in SIM whereas B, the theCPI CPI initially in SIMs would B–D move would up decrease from 2022 by to 0.158%, 2028. Afterwards, 0.155%, and 0.143% on average, respectively. it would diminish slightly and relatively move as constant. Similarly, the CPIs of both SIMs• PrivateC and B Consumption behave similarly. From 2022 to 2027, the CPI would increase, but from 2027 Energies 2021, 14, x FOR PEER REVIEW 13 of 23 to 2031,Figure it would8 reveals be constant the effects. SIM on D privateshows an consumption, increase in CPI which for areonly similar three years to the (2022 GDP– 2025),results. and The then results it would show be thatstable increased and decelerate biofuel until prices 2031, resulting because from the TFP abolishing improvement biofuel andsubsidies investments (SIM A)in energy would crop lead plantations to a decline wo inuld the increase PCON production over the period. output Initially, and reduce the theremovalremoval producer ofof biofuelbiofuel prices subsidiessubsidies of commodities wouldwould affect.affect Therefore, thethe higherhigher the consumer productionproduction prices costscosts andofand finished lowerlower outputs outputsgoods wouldandand wouldwould decrease also also lead accordinglylead to to lower lower valueand value addedpromoting added along alongthe with TFP with a household and a householdinvestment income income reduction.in energy reduction Givencrop . plantationstheGiven decrease the bydecrease in reallocating household in household the income, biofuel people subsidyincome, must arepeople reduce appropriate must their consumption;reduceregimes theirto reduce therefore,consumption; inflation the. householdtherefore,In summary the would household (see adjust Table consumption would 2), during adjust such to consumption satisfy a period, budget the to CPI constraints. satisfy in SIM budget A The would averageconstraints increase PCON. Theby 0wouldaverage.171% on decrease PCON average, would by whereas 0.416% decrease (see the TableCPI by 0in2.416%). SIMs (see B– DTable would 2). decrease by 0.158%, 0.155%, and 0.143% on average, respectively. • Private Consumption Figure 8 reveals the effects on private consumption, which are similar to the GDP results. The results show that increased biofuel prices resulting from abolishing biofuel subsidies (SIM A) would lead to a decline in the PCON over the period. Initially, the

FigureFigure 8.8. ProjectedProjected impactsimpacts ofof fourfour scenariosscenarios (SIMs (SIMs A–D) A–D) on on private private consumption consumption during during 2022–2031. 2022– 2031. To enhance the TFP of energy crops (SIM B), although the average PCON decreased by 0.403%,To enhance it was the still TFP better of energy than in crops SIM (SIM A, because B), although increasing the average the TFP PCON of energy decreased crops wouldby 0.403%, enhance it was the still total better aggregate than in outputs.SIM A, because Therefore, increasing the household the TFP plantedof energy energy crops cropswould would enhance increasingly the total earnaggregate more incomeoutputs together. Therefore, with the consumption. household Inplanted addition, energy the highercrops would productivity increasingly would earn affect more the increaseincome together in outputs with of energyconsumption crops (e.g.,. In addition, sugar cane, the cassava, and oil palm), which are feedstocks of biofuel production. An increase in outputs higher productivity would affect the increase in outputs of energy crops (e.g., sugar cane, would lead to a decrease in producer prices through the biofuel supply chain. Then, the cassava, and oil palm), which are feedstocks of biofuel production. An increase in outputs commercial biofuel prices would decrease depending on the market mechanism, causing would lead to a decrease in producer prices through the biofuel supply chain. Then, the commercial biofuel prices would decrease depending on the market mechanism, causing the cost of other production sectors to also decrease. Therefore, the household could consume more than in SIM A; however, enhancing the TFP of energy crops can partially entail adverse effects. The influence of biofuel subsidy removal would be to substantially affect private consumption. In SIM C, this scenario was to increase the capital supply by reallocating budgets for investment in energy crop plantations. The results showed that the average PCON decreased by 0.403%, equivalent to in SIM B, because the reallocation, in 2022, leads to an increase in capital accumulation, which would result in higher production in the next year. Consequently, the household would gradually receive more income, while the costs of production would inversely decrease. Hence, the PCON and SIM B would increase. Finally, enhancing the TFP along with investment in energy crop plantations (SIM D) showed that both measures considerably enhanced the PCON growth. Although the average consumption would decline, on average, by 0.390%, the trend would be in an incline direction. The result also displayed an impact almost double that in SIMs B and C. Consequently, this implies that the most appropriate option for maintaining the PCON is TFP improvement and reallocation of budget to invest in energy crop plantations. • Gross Fixed Capital Formation Figure 9 demonstrates the changes in the GFCF. The trend of each scenario is different. The oil fund is an organization under government control. Given that biofuel subsidies are cut off, and the government does not need to compensate for biofuel. In SIM A, removing biofuel subsidies could save the oil fund’s liquidation, which would also lead to higher government income and savings. In this study, the total investment is fixed, as

Energies 2021, 14, 2272 13 of 21

the cost of other production sectors to also decrease. Therefore, the household could consume more than in SIM A; however, enhancing the TFP of energy crops can partially entail adverse effects. The inﬂuence of biofuel subsidy removal would be to substantially affect private consumption. In SIM C, this scenario was to increase the capital supply by reallocating budgets for investment in energy crop plantations. The results showed that the average PCON decreased by 0.403%, equivalent to in SIM B, because the reallocation, in 2022, leads to an increase in capital accumulation, which would result in higher production in the next year. Consequently, the household would gradually receive more income, while the costs of production would inversely decrease. Hence, the PCON and SIM B would increase. Finally, enhancing the TFP along with investment in energy crop plantations (SIM D) showed that both measures considerably enhanced the PCON growth. Although the average consumption would decline, on average, by 0.390%, the trend would be in an incline direction. The result also displayed an impact almost double that in SIMs B and C. Consequently, this implies that the most appropriate option for maintaining the PCON is TFP improvement and reallocation of budget to invest in energy crop plantations. • Gross Fixed Capital Formation Figure9 demonstrates the changes in the GFCF. The trend of each scenario is different. The oil fund is an organization under government control. Given that biofuel subsidies are Energies 2021, 14, x FOR PEER REVIEW 14 of 23 cut off, and the government does not need to compensate for biofuel. In SIM A, removing biofuel subsidies could save the oil fund’s liquidation, which would also lead to higher government income and savings. In this study, the total investment is ﬁxed, as mentioned inmentioned Section 3.4 in. TheSection higher 3.4 prices. The ofhigher commodities prices of caused commodities by increasing caused production by increasing costs wouldproduction affect costs the higher would values affect ofthe stock higher changes. values Owing of stock to thechanges inverse. Owing variations to the between inverse thevariations GFCF and between stock the changes, GFCF the and GFCF stock would changes, thus the decrease. GFCF would thus decrease.

FigureFigure 9.9. ProjectedProjected impacts of four scenarios (SIMs (SIMs A A–D)–D) on on gross gross fixed ﬁxed capital capital formation formation during during – 2022–2031.2022 2031.

InIn SIMSIM B,B, thethe TFPTFP improvementimprovement ofof energyenergy cropscrops leadsleads toto anan increaseincrease inin thethe GFCF,GFCF, becausebecause thethe lowerlower productionproduction costscosts ofof energyenergy cropscrops andand biofuelbiofuel productsproducts affectedaffected lowerlower commoditiescommodities prices,prices, leadingleading toto aa decreasedecrease in in the the values values of of stock stock changes. changes. Then, Then, thethe GFCFGFCF wouldwould increaseincrease directly.directly. InIn SIMSIM C,C, the reallocation of of budget budget to to invest invest in in energy energy crop crop plantations plantations leads leads to toan anincrease increase in inGFCF GFCF.. For For the the same same reason reason as as in in SIM SIM B, B, the the investment in energyenergy cropcrop plantations increased biofuel feedstocks’ outputs and reduced biofuel prices. Given the plantations increased biofuel feedstocks’ outputs and reduced biofuel prices. Given the decrease in production costs and commodity prices, the values of stock changes would decrease in production costs and commodity prices, the values of stock changes would reduce. It would ﬁnally lead to an increase in the GFCF. The trend of the GFCF in SIM C reduce. It would finally lead to an increase in the GFCF. The trend of the GFCF in SIM C was quite different from in SIM B, which implies that reallocating subsidy for investment was quite different from in SIM B, which implies that reallocating subsidy for investment has more inﬂuence on the GFCF than enhancing TFP. has more influence on the GFCF than enhancing TFP. Finally, SIM D combined the TFP with reallocation measures. Although the average GFCF remained less than zero, i.e., 0.018% (see Table 2), the trend increased more than the other scenarios, because of the decrease in commodity prices caused by higher productivity and capital supply of energy crop plantations. In summary, removing biofuel subsidies vastly affects the GFCF whether using TFP improvement or reallocation of subsidy budget; however, both measures are a desirable choice in the long term because the trends are likely to increase. If policymakers combine both measures to implement energy crops, the adverse effects would be rapidly eliminated more than applying only an individual measure.

4.3. Sectoral Impacts Table 3 demonstrates the top five impacts on the output of production sectors. Given that the biofuel subsidy was removed, the effects would directly damage the energy and transportation sectors. In SIM A, the top five sectors, namely ethanol production (cassava-based), coastal and transportation, public road transportation, petroleum refinery, and biodiesel, would decline by 2.874%, 2.396%, 1.800%, 1.722%, and 1.722%, respectively, because a rise in biofuel prices leads to a decrease in biofuel consumption. Therefore, petroleum refineries would decrease production volumes to satisfy domestic demand, and therefore affect the feedstock industry, in particular, ethanol (cassava-based) and biodiesel (B100) productions.

Energies 2021, 14, 2272 14 of 21

Finally, SIM D combined the TFP with reallocation measures. Although the average GFCF remained less than zero, i.e., 0.018% (see Table2), the trend increased more than the other scenarios, because of the decrease in commodity prices caused by higher productivity and capital supply of energy crop plantations. In summary, removing biofuel subsidies vastly affects the GFCF whether using TFP improvement or reallocation of subsidy budget; however, both measures are a desirable choice in the long term because the trends are likely to increase. If policymakers combine both measures to implement energy crops, the adverse effects would be rapidly eliminated more than applying only an individual measure.

4.3. Sectoral Impacts Table3 demonstrates the top five impacts on the output of production sectors. Given that the biofuel subsidy was removed, the effects would directly damage the energy and transportation sectors. In SIM A, the top five sectors, namely ethanol production (cassava-based), coastal and transportation, public road transportation, petroleum refinery, and biodiesel, would decline by 2.874%, 2.396%, 1.800%, 1.722%, and 1.722%, respectively, because a rise in biofuel prices leads to a decrease in biofuel consumption. Therefore, petroleum refineries would decrease production volumes to satisfy domestic demand, and therefore affect the feedstock industry, in particular, ethanol (cassava-based) and biodiesel (B100) productions.

Table 3. Sectoral impacts for SIMs A–D (average during 2022–2031).

Scenario Positive Output % Change Negative Output % Change Industrial electrical machinery 0.296 Ethanol (cassava-based) −2.874 Manufacturing 0.295 Coastal and water transportation −2.396 SIM A Metal and non-metal 0.268 Public road transportation −1.800 Public administration 0.230 Petroleum refinery −1.722 Chemical and textile 0.212 Biodiesel production −1.722 Ethanol (cassava-based) 1.724 Ethanol (molasses-based) −2.676 Sugar cane cultivation 0.558 Coastal and water transportation −2.370 SIM B Sugar refinery 0.551 Public road transportation −1.777 Cassava cultivation 0.316 Petroleum refinery −1.659 Manufacturing 0.292 Biodiesel production −1.659 Ethanol (cassava-based) 1.477 Ethanol (molasses-based) −2.728 Sugar cane cultivation 0.334 Coastal and water transportation −2.356 SIM C Sugar refinery 0.326 Public road transportation −1.765 Manufacturing 0.302 Petroleum refinery −1.640 Cassava cultivation 0.300 Biodiesel production −1.640 Ethanol (cassava-based) 6.324 Ethanol (molasses-based) −4.803 Crude palm oil 1.154 Coastal and water transportation −2.332 SIM D Sugar cane cultivation 0.920 Public road transportation −1.743 Oil palm cultivation 0.910 Petroleum refinery −1.578 Sugar refinery 0.883 Biodiesel production −1.578 Source, authors’ simulations.

In SIM B, the termination of the subsidy was simultaneously implemented with an increase in energy crop’s cultivation efﬁciency and induced the substitution between the main feedstocks of ethanol production. Speciﬁcally, driven by their different costs, the Energies 2021, 14, 2272 15 of 21

cassava is used as a substitute for molasses and becomes the main feedstock. Still, in this case, increased prices of fuels reduce transportation activities together with lowering demands for fossil-based fuels and biodiesel. The simulation results of SIM C show that the policy mix of fuel subsidy removal and a new proportion of energy crop investment yields sectoral impacts similar to those in SIM B. According to the cost gap of ethanol’s feedstocks, this policy mix inﬂuences the substitution between cassava and molasses in ethanol production, and simultaneously the productions of petroleum and biodiesel subside. The outcomes of the last simulation scenario, imposing all combinations of SIMs B and C, indicate that the substitution between ethanol’s feedstocks is the major conse- quence. Similar to previous simulation results, transportation and petroleum production are negatively affected. Corresponding to the higher biofuel prices and increasing crops’ productivity, the outputs of food-related sectors, which are sugar reﬁnery and crude palm oil, increase. With the higher retail price and the lower transportation demand, biodiesel production still declines as compared with that of the base case.

4.4. Sensitivity Analysis Figure 10 shows the results obtained from the Monte Carlo simulation. Following the method applied in previous studies [50,51], the coefficients of elasticity of substitution were jointly randomized based on the assumption normal distribution with the standard deviation (SD) of 10% deviation from the original value. The band of two standard deviations (2SD) have been plotted to identify the boundary of the statistical distribution of repeatedly randomized simulations. Similar to the findings of Puttanapong et al. (2015) [52], in the cases of GDP and private consumption, the bands of 2SD will expand over time, implying that these simulated results can be increasingly deviating from the baseline. On the other hand, in the cases of consumer price and gross fixed capital formation, the bands of 2SD will be narrow over the years, indicating that the variation of simulation outcomes will be lower. Hence, in the dynamic dimension, the variation of the coefficients of elasticity of substitution can lead to both cumulative and tapering effects.

4.5. Limitations The limitations of this study are threefold. First, this study used the recursive dynamic mechanism as the main intertemporal process of capital accumulation. This speciﬁcation can be extended to incorporate the fully dynamic structure incorporating all engonized intertemporal adjustments [31,32]. Second, the knowledge spillover process was arbitrarily assigned in SIM D, but this relationship can be endogenized in the future study [53–55]. Third, the model constructed in this study did not include the connection between the adjustment of relative prices and its inﬂuence on technical change. This interrelation can be incorporated in the extended version of the existing CGE model [32,56]. EnergiesEnergies 2021,2021 14, ,x14 FOR, 2272 PEER REVIEW 16 of17 21 of 23

SIM A SIM B

SIM C SIM D

FigureFigure 10. 10. StatisticalStatistical distribution distribution of macroeconomic indices. indices.

Energies 2021, 14, 2272 17 of 21

5. Conclusions and Recommendations In this study, we examine the possible combinations of policies to mitigate the impacts of biofuel subsidy removal, develop a reclusive dynamic CGE model, and apply it as the main tool to conduct simulations, revealing the economy-wide propagation mechanisms of impacts and adjustments. We validate the model’s predicted outcomes with actual 2015–2019 data. The obtained values of the RMSE range from 2.62% to 5.97%, affirming the reliability of the model’s predictive power. We simulate four policy scenarios. The first scenario simulation shows that only terminating the biofuel subsidy causes an economy-wide contraction due to causing higher fuel prices. In addition to the subsidy termination, the second and third scenarios incorporate the improvement of energy crop productivity and reallocation of capital investment in energy crop plantations, respectively. The simulation results identify that policy options can mitigate the negative impacts on the Thai economy. The last simulation, which is the combination of the second and third policy options, yields the lowest economy-wide impacts. These simulation results suggest two recommendations for policymakers. First, termination of the biofuel subsidy along with productivity improvement and reallocation for capital investment in energy crop plantations is the optimal option. Second, substitution between main feedstocks of ethanol production would benefit cassava-based producers and would conversely lower the demand for molasses-based output. Given that the production adjustment requires long-term capital investment, early preparation would mitigate the negative effects on private sectors and consumers.

Author Contributions: Conceptualization, K.P., N.P., and M.P.; data curation, K.P.; formal analysis, K.P.; funding acquisition, M.P.; investigation, K.P. and N.P.; methodology, K.P. and N.P.; project administration, N.P. and M.P.; resource, K.P. and N.P.; software, N.P. and M.P.; supervision, N.P. and M.P.; validation, K.P., N.P., and M.P.; visualization, K.P. and N.P.; writing—original draft preparation, K.P.; writing—review and editing, N.P. and M.P. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the Joint Graduated School of Energy and Environment (JGSEE) and the Sirindhorn International Institute of Technology (SIIT). Conﬂicts of Interest: The authors declare no conﬂict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Appendix A

Table A1. Sectors and commodities in the CGE model

Sector Commodity Industries IO Code Commodities Number Number 001–003, 005–008, 010, 1 1 Agriculture forest and fishery 012–029 Agriculture forest and fishery 2 Cassava plantation 004 2 Cassava 3 Sugarcane plantation 009 3 Sugarcane 4 Oil palm plantation 011 4 Oil palm 5 Food industry 042–054, 056–066 5 Food products 6 Crude palm oil production 047 6 Crude palm oil 7 Sugar refinery 055 7 Sugar products 8 Molasses 8 Coal and lignite mining 030 9 Coal and lignite 9 Crude oil and natural gas 031 10 Crude oil and natural gas 10 Other mining 032–041 11 Other mineral products 11 Petroleum refinery 093–094 12 Liquefied petroleum gas 13 Jet and kerosene 14 Fuel oil 15 Other petroleum products 16 Gasohol E10 and gasoline 17 Gasohol E20 and E85 18 Biodiesel B5/7 (mandatory) 19 Biodiesel B10/B20 (option) Energies 2021, 14, 2272 18 of 21

Table A1. Cont.

Sector Commodity Industries IO Code Commodities Number Number 12 Biodiesel production 093 20 Puriﬁed biodiesel (B100) 13 Ethanol (cassava based) 093 21 Ethanol 14 Ethanol (molasses based) 093 15 Electricity production 135 22 Electricity 16 Natural gas separation 136 23 Natural gas products 17 Metal and non-metal 110–111 24 Metal and non-metal 18 Chemical production 067–092 25 Chemical products 19 Rubber and plastic production 095–109 26 Rubber, plastics, and material 20 Electrical machinery production 112–122 27 Electrical machinery and equipment 21 Transport industry 123–128 28 Transport machinery and maintenance 22 Other manufacturing 129–134 29 Other industrial products 23 Construction 137–144 30 Construction 24 Trading and services 145–148, 160–164 31 Trading and services 25 Public administration 165–171 32 Public administration 26 Railway freight transportation 149 33 Railway freight transportation 27 Railway mass transportation 149 34 Railway mass transportation 28 Road public transportation 150 35 Road public transportation 29 Road freight transportation (Heavy) 151 36 Road freight transportation (Heavy) 30 Road freight transportation (Light) 151 37 Road freight transportation (Light) 31 Land transportation services 152 38 Land transportation services 32 Ocean and coastal transportation 153–154 39 Ocean and coastal transportation 33 Water transportation services 155 40 Water transportation services 34 Air transportation 156 41 Air transportation 35 Other services and activities 157–159, 172–180 42 Other services and activities

Table A2. Coefﬁcients of elasticity of substitution.

Coefficients of Elasticity of Substitution Sector Number Industries Biodiesel (B5 vs. Gasohol (E10 vs. Labor vs. Capital [a] B7/B10) [b] E20/E85) [b] 1 Agriculture forest and fishery 1.15 0.5 0.5 2 Cassava plantation 1.15 0.5 0.5 3 Sugarcane plantation 1.15 0.5 0.5 4 Oil palm plantation 1.15 0.5 0.5 5 Food industry 0.86 0.5 0.5 6 Crude palm oil production 0.86 0.5 0.5 7 Sugar refinery 0.86 0.5 0.5 8 Coal and lignite mining 1.1 0.5 0.5 9 Crude oil and natural gas 1.1 0.5 0.5 10 Other mining 1.1 0.5 0.5 11 Petroleum refinery 1.1 0.5 0.5 12 Biodiesel production 1.1 0.5 0.5 13 Ethanol (cassava based) 0.86 0.5 0.5 14 Ethanol (molasses based) 0.86 0.5 0.5 15 Electricity production 1.02 0.5 0.5 16 Natural gas separation 1.02 0.5 0.5 17 Metal and non-metal 0.86 0.5 0.5 18 Chemical production 0.86 0.5 0.5 19 Rubber and plastic production 0.86 0.5 0.5 20 Electrical machinery production 0.86 0.5 0.5 21 Transport industry 0.87 0.5 0.5 22 Other manufacturing 0.86 0.5 0.5 23 Construction 0.97 0.5 0.5 24 Trading and services 0.87 0.5 0.5 25 Public administration 1.04 0.5 0.5 26 Railway freight transportation 1.03 0.5 0.5 27 Railway mass transportation 1.03 0.5 0.5 28 Road public transportation 1.03 0.5 0.5 29 Road freight transportation (Heavy) 1.03 0.5 0.5 30 Road freight transportation (Light) 1.03 0.5 0.5 Energies 2021, 14, 2272 19 of 21

Table A2. Cont.

Coefﬁcients of Elasticity of Substitution Sector Number Industries Biodiesel (B5 vs. Gasohol (E10 vs. Labor vs. Capital [a] B7/B10) [b] E20/E85) [b] 31 Land transportation services 1.03 0.5 0.5 32 Ocean and coastal transportation 1.03 0.5 0.5 33 Water transportation services 1.03 0.5 0.5 34 Air transportation 1.03 0.5 0.5 35 Other services and activities 1.03 0.5 0.5 Notes: [a] is based on [37]; [b] is based on assumption of authors; For the other types of elasticity i.e., the elasticity of transformation between products of industry j = 2, the elasticity of transformation between exports and local supplies of commodity i by industry j = 2, and the elasticity of substitution between imported and domestically produced commodity i = 2, are based on [24].

Table A3. Dynamic parameters.

Year n eg ig itg etr ppg δ γ 2019–2031 0.01 0.05 0.05 0.055 0.05 0.01 0.07 2 Notes: n = population growth [33]; eg = government expenditures growth [34]; ig = public sector investment growth [34]; itg = total investment growth [34]; etr = export growth [34]; ppg = total factor productivity growth [35]; δ = depreciation rate [36,37]; γ = elasticity of investment demand [36,37].

References 1. Asian Development Bank. Fossil Fuel Subsidies in Thailand: Trends, Impacts, and Reforms; Asian Development Bank: Manila, Philippines, 2015. 2. Wattana, S. Bioenergy Development in Thailand: Challenges and Strategies. Energy Procedia 2014, 52, 506–515. [CrossRef] 3. Thai Ethanol Association. Reference Ethanol Prices. Available online: http://www.thai-ethanol.com/th/statistical-data/price. html (accessed on 27 February 2021). 4. Department of Alternative Energy Development and Efficiency. Biodiesel Price and Producer Data. Available online: https: //www.dede.go.th/more_news.php?cid=88&filename=index (accessed on 27 February 2021). 5. Chanthawong, A.; Dhakal, S.; Kuwornu, J.K.M.; Farooq, M.K. Impact of Subsidy and Taxation Related to Biofuels Policies on the Economy of Thailand: A Dynamic CGE Modelling Approach. Waste Biomass Valorization 2020, 11, 909–929. [CrossRef] 6. United States Department of Agriculture. Thailand Biofuels Annual. Available online: https://apps.fas.usda.gov/newgainapi/ api/Report/DownloadReportByFileName?fileName=Biofuels%20Annual_Bangkok_Thailand_11-04-2019 (accessed on 5 Febru- ary 2021). 7. Office of Energy and Policy Planning. Oil Prices. Available online: http://www.eppo.go.th/epposite/index.php/th/petroleum/ price/oil-price?category_id=558&isc=0&orders[publishUp]=publishUp&issearch=1 (accessed on 5 February 2021). 8. Koesler, S.; Pothen, F. The Basic WIOD CGE Model: A Computable General Equilibrium Model Based on the World Input-Output Database; Zentrum für Europäische Wirtschaftsforschung: Mannheim, Germany, 2013. 9. Asian Productivity Organization. APO Productivity Databook 2020; Asian Productivity Organization: Tokyo, Japan, 2020. 10. Warr, P.; Suphannachart, W. Agricultural Productivity Growth and Poverty Reduction: Evidence from Thailand; Departmental Working Papers; The Australian National University, Arndt-Corden Department of Economics: Canberra, Australia, 2020. 11. Fattouh, B.; El-Katiri, L. Energy Subsidies in the Middle East and North Africa. Energy Strategy Rev. 2013, 2, 108–115. [CrossRef] 12. Bazilian, M.; Onyeji, I. Fossil Fuel Subsidy Removal and Inadequate Public Power Supply: Implications for Businesses. Energy Policy 2012, 45, 1–5. [CrossRef] 13. Solaymani, S.; Kari, F. Impacts of Energy Subsidy Reform on the Malaysian Economy and Transportation Sector. Energy Policy 2014, 70, 115–125. [CrossRef] 14. Breisinger, C.; Engelke, W.; Ecker, O. Leveraging Fuel Subsidy Reform for Transition in Yemen. Sustainability 2012, 4, 2862–2887. [CrossRef] 15. Lin, B.; Li, A. Impacts of Removing Fossil Fuel Subsidies on China: How Large and How to Mitigate? Energy 2012, 44, 741–749. [CrossRef] 16. Lin, B.; Jiang, Z. Estimates of Energy Subsidies in China and Impact of Energy Subsidy Reform. Energy Econ. 2011, 33, 273–283. [CrossRef] 17. Liu, W.; Li, H. Improving Energy Consumption Structure: A Comprehensive Assessment of Fossil Energy Subsidies Reform in China. Energy Policy 2011, 39, 4134–4143. [CrossRef] 18. Preecha, R.; Wianwiwat, S. The Effect of Abolishing the Oil Fund on the Thai Economy: A Computable General Equilibrium Analysis. Int. Energy J. 2017, 17, 155–162. Energies 2021, 14, 2272 20 of 21

19. Bridle, R.; Sharma, S.; Mostafa, M.; Geddes, A. Fossil Fuel to Clean Energy Subsidy Swaps; International Institute for Sustainable Development: Winnipeg, MB, Canada, 2019. 20. Castel, V. Economic Brief-Reforming Energy Subsidies in Egypt. Available online: https://www.afdb.org/en/documents/ document/economic-brief-reforming-energy-subsidies-in-egypt-26669 (accessed on 13 April 2021). 21. Hartono, D.; Komarulzaman, A.; Irawan, T.; Nugroho, A. Phasing out Energy Subsidies to Improve Energy Mix: A Dead End. Energies 2020, 13, 2281. [CrossRef] 22. Henseler, M.; Maisonnave, H. Low World Oil Prices: A Chance to Reform Fuel Subsidies and Promote Public Transport? A Case Study for South Africa. Transp. Res. Part Policy Pract. 2018, 108, 45–62. [CrossRef] 23. Farajzadeh, Z.; Bakhshoodeh, M. Economic and Environmental Analyses of Iranian Energy Subsidy Reform Using Computable General Equilibrium (CGE) Model. Energy Sustain. Dev. 2015, 27, 147–154. [CrossRef] 24. Decaluwe, B.; Lemelin, A.; Maisonnave, H.; Robichaud, V. PEP Standard CGE Models. Available online: https://www.pep-net. org/pep-standard-cge-models (accessed on 17 July 2020). 25. Office of the National Economic and Social Development Council. Input-Output Tables (I-O Tables). Available online: https: //www.nesdc.go.th/nesdb_en/more_news.php?cid=158&filename=index (accessed on 19 July 2020). 26. Department of Alternative Energy Development and Efficiency. Energy Statistics. Available online: https://www.dede.go.th/ ewt_news.php?nid=42079 (accessed on 17 May 2020). 27. Department of Energy Business, Ministry of Energy. Statistic. Available online: https://www.doeb.go.th/2017/EN_index.html? ln=en#/article/en_statistic (accessed on 10 March 2021). 28. Bank of Thailand. Economics Statistics. Available online: https://www.bot.or.th/Thai/Statistics/EconomicAndFinancial/ RealSector/Pages/Index.aspx (accessed on 17 May 2020). 29. Office of Agricultural Economics. Agricultural Statistics of Thailand 2018. Available online: http://www.oae.go.th/assets/ portals/1/files/jounal/2563/yearbook62edit.pdf (accessed on 18 May 2020). 30. The Thailand Research Fund E-Library. Available online: https://elibrary.trf.or.th/project_content.asp?PJID=RDG5150022 (accessed on 11 February 2021). 31. Bretschger, L.; Ramer, R.; Schwark, F. Growth Effects of Carbon Policies: Applying a Fully Dynamic CGE Model with Heteroge- neous Capital. Resour. Energy Econ. 2011, 33, 963–980. [CrossRef] 32. Bretschger, L.; Zhang, L. Carbon Policy in a High-Growth Economy: The Case of China. Resour. Energy Econ. 2017, 47, 1–19. [CrossRef] 33. Organisation for Economic Co-operation and Development (OECD); International Labor Organization (ILO). How Immigrants Contribute to Thailand’s Economy; OECD Publishing: Paris, France, 2017. 34. Office of the National Economic and Social Development Council National Income of Thailand. Available online: https: //www.nesdc.go.th/nesdb_en/more_news.php?cid=154&filename=index (accessed on 11 February 2021). 35. University of Groningen; University of California, Davis Total Factor Productivity at Constant National Prices for Thailand. Available online: https://fred.stlouisfed.org/series/RTFPNATHA632NRUG (accessed on 10 March 2021). 36. Annabi, N.; Cockburn, J.; Decaluwe, B. A Sequential Dynamic CGE Model for Poverty Analysis. Available online: http://www. pep-net.org/sites/pep-net.org/files/typo3doc/pdf/files_events/3rd_dakar/SeqDynCGE.pdf. (accessed on 15 March 2021). 37. Kaenchan, P.; Puttanapong, N.; Bowonthumrongchai, T.; Limskul, K.; Gheewala, S.H. Macroeconomic Modeling for Assessing Sustainability of Bioethanol Production in Thailand. Energy Policy 2019, 127, 361–373. [CrossRef] 38. Hosoe, N.; Gasawa, K.; Hashimoto, H. Textbook of Computable General Equilibrium Modeling: Programming and Simulations, 2010th ed.; Palgrave Macmillan: Basingstoke, UK, 2010; ISBN 978-0-230-24814-4. 39. Suphannachart, W.; Warr, P. Total factor productivity in Thai agriculture: Measurement and determinants. In Productivity Growth in Agriculture: An International Perspective; Fuglie, K.O., Wang, S.L., Ball, V.E., Eds.; CABI: Wallingford, UK, 2012; pp. 215–236. ISBN 978-1-84593-921-2. 40. Mary, S.; Phimister, E.; Roberts, D.; Santini, F. A Monte Carlo Filtering Application for Systematic Sensitivity Analysis of Computable General Equilibrium Results. Econ. Syst. Res. 2019, 31, 404–422. [CrossRef] 41. Belgodere, A.; Vellutini, C. Identifying Key Elasticities in a CGE Model: A Monte Carlo Approach. Appl. Econ. Lett. 2011, 18, 1619–1622. [CrossRef] 42. Acharya, R.H.; Sadath, A.C. Implications of Energy Subsidy Reform in India. Energy Policy 2017, 102, 453–462. [CrossRef] 43. Moshiri, S.; Martinez Santillan, M.A. The Welfare Effects of Energy Price Changes Due to Energy Market Reform in Mexico. Energy Policy 2018, 113, 663–672. [CrossRef] 44. Aune, F.R.; Grimsrud, K.; Lindholt, L.; Rosendahl, K.E.; Storrøsten, H.B. Oil Consumption Subsidy Removal in OPEC and Other Non-OECD Countries. Oil Market Impacts and Welfare Effects; Discussion Papers; Statistics Norway, Research Department: Oslo, Norway, 2016. 45. Bhattacharyya, R.; Ganguly, A. Cross Subsidy Removal in Electricity Pricing in India. Energy Policy 2017, 100, 181–190. [CrossRef] 46. Wang, Y.; Ali Almazrooei, S.; Kapsalyamova, Z.; Diabat, A.; Tsai, I.-T. Utility Subsidy Reform in Abu Dhabi: A Review and a Computable General Equilibrium Analysis. Renew. Sustain. Energy Rev. 2016, 55, 1352–1362. [CrossRef] 47. Wesseh, P.K.; Lin, B.; Atsagli, P. Environmental and Welfare Assessment of Fossil-Fuels Subsidies Removal: A Computable General Equilibrium Analysis for Ghana. Energy 2016, 116, 1172–1179. [CrossRef] Energies 2021, 14, 2272 21 of 21

48. AlDarraji, H.H.M.; Bakir, A. The Impact of Renewable Energy Investment on Economic Growth. J. Soc. Sci. COESRJ-JSS 2020, 9, 234–248. [CrossRef] 49. Karydas, C.; Zhang, L. Green Tax Reform, Endogenous Innovation and the Growth Dividend. J. Environ. Econ. Manag. 2019, 97, 158–181. [CrossRef] 50. Puttanapong, N. The Computable General Equilibrium (CGE) Models with Monte-Carlo Simulation for Thailand: An Impact Analysis of Baht Appreciation. Ph.D Dissertation, Cornell University, Ithaca, NY, USA, 2008. 51. Puttanapong, N. Impacts of Climate Change on Major Crop Yield and the Thai Economy: A Nationwide Analysis Using Static and Monte-Carlo Computable General Equilibrium Models. Thail. World Econ. 2013, 31, 68–87. 52. Puttanapong, N.; Wachirarangsrikul, S.; Phonpho, W.; Raksakulkarn, V. A Monte-Carlo Dynamic CGE Model for the Impact Analysis of Thailand’s Carbon Tax Policies. J. Sustain. Energy Environ. 2015, 6, 43–53. 53. Aghion, P.; Dechezleprêtre, A.; Hémous, D.; Martin, R.; Van Reenen, J. Carbon Taxes, Path Dependency, and Directed Technical Change: Evidence from the Auto Industry. J. Polit. Econ. 2016, 124, 1–51. [CrossRef] 54. Calel, R.; Dechezleprêtre, A. Environmental Policy and Directed Technological Change: Evidence from the European Carbon Market. Rev. Econ. Stat. 2016, 98, 173–191. [CrossRef] 55. Dechezleprêtre, A.; Martin, R.; Mohnen, M. Knowledge Spillovers from Clean and Dirty Technologies; Centre for Climate Change Economics and Policy, London School of Economics and Political Science: London, UK, 2014. 56. Popp, D. Induced Innovation and Energy Prices. Am. Econ. Rev. 2002, 92, 160–180. [CrossRef]